1. Field of the Invention
The present invention relates to an evaluation method, apparatus, and recording medium using an optimum template pattern determination method, apparatus, and an optimum template pattern, used in, for example, correcting and evaluating a translation. In particular, the present invention relates to an evaluation method, apparatus, and recording medium using an optimum pattern determination method, apparatus, and an optimum template pattern which allow for determining an optimum template pattern based on some words having greater importance than other words and evaluating an input sentence based on the optimum template pattern.
2. Description of the Prior Art
In technical translator training courses, a method is generally used in which students send their assigned translations to the tutoring center, where the translations are corrected, evaluated, and sent back to the students, and the students review the corrections and know their rating.
Conventionally, a method has been used to correct translations in which a large number of translations sent from students are distributed among a number of tutors and each of the tutors manually corrects errors in the translations. However, there are problems with this method: the manual correction requires a great deal of time. In addition, since it is not necessarily easy to employ tutors having ability at a certain level or higher, the corrections may vary depending on the tutors.
The inventors previously proposed a translation correction support apparatus that provides results nearly equal to those from manual correction by experts, as disclosed in Japanese Patent Laid-Open No. 9-325673.
In the translation correction support apparatus previously proposed by the inventors, a plurality of template patterns are provided, each of which corresponds to a model translation. A template pattern which matches the student""s translation is determined, the translation is regarded in the light of templates similar to the matching template pattern, then the differences between each template and the student""s translation are determined to produce the resulting corrections. Very precise corrections can be obtained by this apparatus if the student""s translations have only few errors.
However, if the student""s translation has a number of errors as in, for example, general language schools, it is impossible for a given model translation, to provide sufficient quantity of corresponding template patterns which accommodate all the possible error variations. As a result, it often occurs that a translation matches none of the translation patterns of the model translation provided beforehand.
Conventionally, in such an application, the student""s translation is compared with each of the template patterns of a model translation and the template pattern having the largest number of words that match words in the student""s translation is selected as the optimum template pattern. However, if the template pattern for the model translation is selected simply because it has the largest number of matching words, the selected template pattern may not necessarily match the template pattern which the student intends in the translation and therefore a precise correction and a proper evaluation of the input translation cannot be expected.
Such a problem occurs not only in the translation correction support apparatus, but also often in information retrieval, in which certain information is retrieved based on query in an information retrieval system using the Internet, for example.
The present invention was devised in view of the above-mentioned state, and it is an object of the present invention to provide an optimum template decision method, apparatus, and recording medium which ensure that the same template pattern as that of an input sentence which the person inputting the sentence intends is selected as the template for a model sentence.
It is another object of the present invention to provide an evaluation method, apparatus, and recording medium using an optimum template pattern which allow a proper evaluation of an input sentence based on the optimum template pattern selected in the above-mentioned way.
To achieve these objects, the present invention provides a method for determining an optimum template pattern for the model sentence, wherein the plurality of template patterns for a model sentence, each of which being regarded as an arbitrary number of templates, are provided beforehand. Each of the template patterns is compared with an input sentence to determine a template pattern most appropriate for the input sentence as the optimum template pattern. This method is characterized by assigning scores to all the words used in each of the template patterns according to their importance and comparing the input sentence with each of the template patterns to determine the template pattern having the highest total of scores of all the matching words as the optimum template pattern. Since the scores are assigned to the words according to their importance and the optimum template pattern is determined based on the total of the scores of matching words, it is ensured that the same template pattern as input sentence""s template pattern intended by the person who input that sentence can be selected and determined as the optimum template pattern for the model sentence.
The present invention further provides an evaluation method using an optimum template pattern, wherein a plurality of template patterns for a model sentence, each of which is regarded as an arbitrary number of templates, are provided beforehand. Each of the template patterns is compared with an input sentence to determine the template pattern most appropriate for the input pattern as an optimum template pattern. Then the input sentence is evaluated based on the optimum template pattern. The method is characterized by assigning scores to all the words used in each of the template patterns according to their importance and comparing the input sentence with each of the template patterns to determine the template pattern having the highest total of scores of all the matching words as the optimum template pattern. Then the input sentence is evaluated based on the product that is yielded by multiplying the number of the matching words in the input sentence, divided by the total number of the words in the input sentence, by the score of the matching words in the input sentence, divided by the total score of the optimum template pattern. Thus, the input sentence can be evaluated properly by taking into account whether the person who input the sentence understands the rules of grammar and knows important words.
The present invention further provides an apparatus for determining an optimum template pattern, wherein a plurality of template patterns for a model sentence, each of which is regarded as an arbitrary number of templates, are provided beforehand. Each of the template patterns is compared with the input sentence to determine the template pattern most appropriate for the input pattern as an optimum template pattern. The apparatus comprises a template pattern storage means for storing each of the template patterns; an input sentence storage means for storing the input sentence; and a word score storage means for storing scores assigned to all the words used in each of the template patterns according to their importance, by associating the scores with the words. The apparatus further comprise a matching word retrieval means for retrieving words in each template pattern which match the words in the input sentence; a score calculation means for calculating the total of the scores of matching words in each template pattern; and an optimum template pattern selection means for selecting the template pattern having the highest total score as the optimum template pattern; wherein the total of the scores of matching words is calculated in said score calculation means and the optimum template pattern is selected based on the result of the calculation. Thus, the likelihood that a template pattern containing a larger number of important words is selected as the optimum template pattern is increased and therefore it is ensured that the same template pattern as the input sentence""s template pattern intended by the person who input that sentence is selected and determined as the optimum template pattern for the model sentence.
The present invention provides an evaluation apparatus using an optimum template pattern, wherein a plurality of template patterns for a model sentence, each of which being regarded as an arbitrary number of templates, are provided beforehand. Each of the template patterns is compared with the input sentence to determine the template pattern most appropriate for the input pattern as an optimum template pattern. Then evaluating the input sentence is evaluated based on the optimum template pattern. The evaluation apparatus comprises a template pattern storage means for storing each of said template patterns; an input sentence storage means for storing said input sentence; and a word score storage means for storing in a memory area the scores assigned to all the words used in each of said template patterns according to their importance by associating the scores with the words. The evaluation apparatus further comprises a matching word retrieval means for retrieving words in each template pattern which match the words in the input sentence; a score calculation means for calculating the total of the scores of matching words in each template pattern; an optimum template pattern selection means for selecting a template pattern having the highest total score as the optimum template pattern; an optimum template pattern total score calculation means for calculating the total score of the optimum template pattern; a matching word count calculation means for calculating the number of matching words in the input sentence; and an input sentence evaluation means for evaluating the input sentence based on the product that is yielded by multiplying the number of the matching words in the input sentence, divided by the number of all the words in the input sentence, by the score of the matching words in the input sentence divided by the total score of the optimum template pattern. Thus, the same template pattern as the input sentence""s template pattern intended by the person who input that sentence can be used to evaluate the input sentence, allowing for high reliability.
The present invention is further characterized by causing a computer to perform processes of: storing in a memory area a plurality of template patterns for a model sentence, each of which is regarded as an arbitrary number of templates; storing an input sentence in a memory area; storing the scores assigned to words used in each of the template patterns according to their importance by associating the scores with the words; and comparing the input sentence with each of the template patterns to select a template pattern having the highest total of the scores of matching words as an optimum template pattern. By the above-mentioned processes, the likelihood that a template pattern containing a larger number of important words having a higher score is selected as the optimum template pattern is increased and therefore the same template pattern as the input sentence""s template pattern intended by the person who input that sentence can be selected and determined as the optimum template pattern for the model sentence.
The present invention is further characterized by causing a computer to perform processes of: storing in a memory area a plurality of template patterns for a model sentence, each of which is regarded as an arbitrary number of templates; storing an input sentence in a memory area; storing in a memory area the scores assigned to the words used in each of the template pattern according to their importance, by associating the scores with the words; comparing the input sentence with each of the template patterns to select the template pattern having the highest total of the scores of matching words as the optimum template pattern. The computer next evaluates the input sentence based on the product that is yielded by multiplying the number of the matching words in the input sentence divided the number of all the words in the input sentence, by the score of the matching words in the input sentence, divided by the total score of the optimum template pattern. By the above-mentioned processes, the input sentence can be evaluated properly.